چكيده به لاتين
ABSTRACT
A semi-empirical and generalizable model of the engine performance is proposed in this thesis whose inputs are engine control variables and ambient air conditions (12 variables), and its output is engine torque. Model development is based on the analysis of the variations of the engine behavior with the variations of its working variables. Furthermore the physical relations governing the engine behavior are also used where possible. Regarding the large number of input variables, and hence the large number of the required tests, the analysis of the engine behavior in different conditions is used by applying a physical comprehensive model of XU7/JPL3 engine.
In the first step, considering the large number of input variables, and in order to simplify the modeling process, the model is broken into several sub-models by use of physical relations, and each sub-model is then developed separately. The final model is produced by combining these sub-models in multiplicative manner. In the next step, in order to reduce the complexity of the proposed model, a systematic approach is used to simplify the model. The result is a model with 58 unknown coefficients that considers a large number engine working variables, needs a small amount of data for regression (determination of the unknown coefficients), and has good accuracy in prediction of unseen data. Furthermore, based on the physical relations, the proposed model is also applicable for prediction of the engine volumetric efficiency.
Finally the proposed model is verified using the numerical and experimental data of seven different engines. Results show that the prediction error is below 10% in all conditions. Furthermore the mean of training and test errors are 3% and 4% for volumetric efficiency prediction, and 3% and 5% for torque prediction, respectively. The accuracy of the proposed model for prediction of different engines behavior in different conditions is a proof of the its accuracy in different regions of engine performance space, and also its generalizability for different engines.
Keywords: Semi-empirical modeling, Spark ignition engine, Engine performance, Torque, Volumetric efficiency
